Abstract
Our project StreetSavvi is an effort to bring the power of AI, Digital Twin technology, advanced simulation and visualization to local Stakeholders such as the City of Philadelphia, Penndot and SEPTA for a partnership during the Route for Change Project which will redevelop the Philadelphia Roosevelt Boulevard.
In an ever-increasing Digital Age, the US DOT has put great effort into developing a large number of databases that encompass infrastructure, travel behavior and safety, and context as a way to answer the formidable challenge of moving people and goods in a safe and efficient way. Interestingly, commercial satellite and street imagery such as those produced by Google and other geospatial data providers emerge as an opportunity for an up-to-date inventory of transportation infrastructure. In addition, the acceleration of Artificial Intelligence through Machine Learning and its inference processes for data processing or the Large Language Models generative processes offer phenomenal opportunities to take a global approach to the field of transportation.
In this context, the team proposes the development of StreetSavvi, an advanced Digital Twin development platform, which applies AI computer vision processes to geospatial data, namely satellite imagery, street imagery, and camera data, to develop photorealistic 3D models of neighborhoods and develop a scalable, efficient, and user-centric digital twin platform for urban transportation planning.
This project expands on a previous Safety21 University Transportation Center feasibility project. It also expands on the installation of speed cameras on the Roosevelt Blvd. Our proposed Streetsavvi project will closely follow the design alternatives developed by PennDOT, the City of Philadelphia, and SEPTA as they include components such as Neighborhood boulevard, Partially-capped expressway, Bus Rapid Transit (BRT), Light Rail Transit (LRT), Subway into the new design. Our simulation and visualization tools will allow the research team to partner with the stakeholders to assess the impact of various designs on multiple metrics such as traffic congestion, safety, cost… Our virtual models will allow an efficient assessment of the proposed infrastructural changes, such as lane removals, enhanced bicycle/pedestrian facilities, and exclusive bus only lanes. We propose with this project to automate the ideation and simulation process by creating a friendly AI based user interface that is easy to use by state, local, and tribal transportation agencies who are stakeholders in the design of our streets.
Description
Timeline
Strategic Description / RD&T
Section left blank until USDOT’s new priorities and RD&T strategic goals are available in Spring 2026.
Deployment Plan
Q1) Stakeholders will be invited for a review of what has been accomplished thus far on the Philadelphia Roosevelt Blvd in the original Safety21 project. The route for Change propositions will be reviewed and a plan developed for priorities for the Digital Twin development. A calendar will be developed to ensure the research team is integrated with the Route for Change townhall reviews.
Q2) The second quarter will be used to develop the alternative routes assigned in Quarter 1 by the stakeholders. A review with stakeholders will be planned at the end of the second quarter.
Q3) The third quarter will be used to extract current traffic flow information from the Roosevelt Speed Cameras and inform a parametrized stochastic traffic model.
Q4) Stakeholders will be invited to review the completed work. They will be invited to visualize the results of the work through a Virtual Reality simulation.
Expected Outcomes/Impacts
Digital twins have a transformative role in transportation by enabling real-time monitoring, simulation, and optimization of transport systems. For transportation infrastructure, digital twins create dynamic models of roads, bridges, and tunnels, allowing cities to simulate traffic flow, identify bottlenecks, and plan improvements based on real-time data. This can lead to smarter traffic management, better route planning, and safer environments for drivers and pedestrians.
Our use case of the Philadelphia Roosevelt Boulevard with a direct participation of stakeholders such as the City of Philadelphia, SEPTA and Penndot will ensure our results are well understood and used by the decision makers who have direct authority over the Route for Change project.
Expected Outputs
The proposed work’s main anticipated outputs include:
- Research Impact: The proposed work will develop novel semi automatic approaches for the development of driveable Digital Twins. This project will integrate various tools such as OpenStreetMap, RoadRunner, Cesium, Unity and Unreal Engine API. The perspective of automating Digital Twin cities roadways generation is a paradigm shift for road and even urban planning with solid impact in transportation.
-Publications: Research developed through the proposed work will be disseminated through publications at major international venues such as IEEE, TRB Annual Conference, TRR Transportation Journal, SAE…
- Data: Our generated data will be integrated to OpenStreetMap database so it can be used as a public database.
- Software: Source code and associated software systems will be developed.
- Student Training: Master and undergraduate students will be trained during the course of the research. The students will actively collaborate with the transportation stakeholders.
- Town Halls demonstration: our software will be made available through demonstrations at the various town halls organized to discuss the Route for Change initiative.
TRID
Our TRiD search using the project keywords resulted in 4 entries (see uploaded file), which all have a date later than 2020, a validation of the novelty of this research field. The first entry relates to the modelization of fires in tunnels. It is a narrow use case of our much wider application: the semi automatic generation of roadways for cities.
The second entry proposes data fusion methodology to integrate camera image and Digital Twin information from the cloud to help intelligent vehicles make smart driving decisions. It does not develop a 3D model of roadways like our proposed research.
The third entry develops ways for connected vehicles to cooperate with each other to cross intersections. This is again a narrow which does not suggest how to precisely develop the Digital Twin itself from satellite imagery. The fourth entry presents a mixed-reality test framework for visualizing sensor feeds the Unity 3D game engine. It is an interesting application that supports the testing of self-driving algorithms. However it focuses on the use of Digital Twin rather than its development. In summary, the proposed work demonstrates an untapped research need for semi automated procedures for Digital Twin development.
Individuals Involved
| Email |
Name |
Affiliation |
Role |
Position |
| xiaoxiad@upenn.ed |
Dong, Xiaoxia |
University of Pennsylvania |
Co-PI |
Other |
| erickg@upenn.edu |
Guerra, Erick |
University of Pennsylvania |
PI |
Other |
| LoebH@email.chop.edu |
Loeb, Helen |
University of Pennsylvania |
Co-PI |
Other |
Budget
Amount of UTC Funds Awarded
$35000.00
Total Project Budget (from all funding sources)
$70000.00
Documents
Match Sources
No match sources!
Partners
| Name |
Type |
| Jitsik |
Deployment Partner_ Deployment Partner_ |
| The Bicycle Coalition of Greater Philadelphia |
None |
| City of Philadelphia |
None |
| Delaware Valley Regional Planning Commission |
None |
| SEPTA |
None |
